Bridge Monitoring Data Reconstruction Study Based on Prophet-Transformer

In order to better meet the needs of bridge monitoring systems, this paper constructs the Prophet-Transformer time-series prediction model method based on the combination of Transformer and Prophet algorithms, for the reconstruction study of missing data, and explores the application characteristics and detailed modeling process of the model in depth. In this paper, the missing deflection data of the bridge is predicted in the context of Meixi River Bridge of Zhengwan Line. Compared with the single Transformer model, the Prophet-Transformer model has higher prediction accuracy, as well as lower MAE and RMSE of 0.0825 and 0.1104, respectively. Experimental results show that the time series obtained by InfluxDB data can effectively recover the missing data of bridges after processing by Prophet-Transformer model, which makes the bridge monitoring data have better analyzability and higher utilization.